Skip to main content.

Appendix 2: NO to NO2 Conversion

This is an update of the conversion methodology given in the Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004a). It is broadly consistent with that methodology, but contains more details, based on more recent analysis of NOx monitoring, and some examples.

Practical methods to estimate nitrogen dioxide concentrations from nitrogen oxide levels

Introduction

Emissions of nitrogen oxides are generally estimated in terms of the composite NOx, made up of nitric oxide and nitrogen dioxide (NO and NO2); ie, NOx = NO + NO2. For most combustion sources, including vehicle emissions and discharges from power stations, the emission will be more than 90% NO, which oxidises quite slowly to NO2 in the atmosphere. The contaminant of interest is NO2, which is included in the national environmental standards for air quality (the Standards) because of its health effects and its potential to degrade visibility. Robust estimates of NO2 concentrations in most cases must be derived from information about emissions of NOx from nearby sources, and several methods for doing this are well established (eg, Cole and Summerhays, 1979).

However, this appendix is aimed at practitioners, or those with responsibility for demonstrating compliance with the Standards for ambient NO2 concentrations. The national ambient air quality standard for NO2 is set at 200 µg/m3 (one-hour average) with nine exceedances allowed per year (Ministry for the Environment, 2004a). In New Zealand, for most sites and most time periods, the NO2 concentrations will be well below the NO2 standard, and the estimates by the methods outlined in this document are all that will be required to demonstrate full compliance. If these methods show possible exceedances of the NO2 standard, then more sophisticated monitoring and modelling will be necessary. This approach is consistent with a Tier 2 process.

Some Auckland monitoring sites are known to have NO2 exceedances (eg, Queen Street, Khyber Pass and Penrose). In other cities, future industrial and roading projects pose a potential NO2 problem. For these cases, where there are known or potential exceedance problems the simple, empirical and conservative methods outlined in this Appendix may not be the most appropriate, and specialised monitoring and interpretation will be required. Here two methods are outlined for practical (but very conservative) estimation of a relationship between NO2 and NOx for the assessment of NO2 compliance from observed or modelled NOx, particularly when monitoring information is sparse or not available and the need for a Tier 2 screening assessment of an emission source is indicated.

If NOx emissions are known, together with the corresponding meteorological information, dispersion models can simulate the spread of NOx as if it were an un-reactive gas (ie, the total number of molecules of NOx does not change through the dispersion and oxidation process). However, the determination of the fraction of this NOx that is NO2 requires a model that either simulates chemical transformations or uses some empirically determined formula for the NO2−NOx relationship. Even when a sophisticated model is available to simulate the oxidation of NO to NO2, knowledge of the oxidants taking part in such reactions (eg, ozone, volatile hydrocarbon products) is still unlikely to be adequate as model input.

Hourly average NOx and NO2 data are used here, but the methods outlined are also suitable for other averaging times (eg, 24 hours).

Relationship between NO2 and NOx measurements

Ambient NOx and NO2 concentrations are routinely monitored at several locations in New Zealand within or close to major urban areas. Figure A2.1 shows scatter plots of NO2 and NOx one-hour concentrations at three different types of urban site in Auckland in 2001, and is typical of other recent years.

  • Khyber Pass is close to a congested urban intersection. This is a problem NO2 site of the sort mentioned above, and needs specialist and detailed analysis rather than the simple approach taken here (which is appropriate for most sites).

  • Takapuna is a typical urban area with commercial and light industrial activity.

  • Musick Point, although not strictly urban, is frequently downwind of the Penrose industrial area and the busy Southern Motorway.

Figure A2.1: Relationship between NO2 and NOx one-hour average concentrations at three Auckland sites in 2001

See figure at its full size (including text description).

For a site like Musick Point near a large city, 9% of all points on the plot in Figure A2.1 are at very low NOx levels (less than 25 µg/m3), with NO2 somewhat less, although the NO2:NOx ratio is often high because the dispersion and oxidation of distant emissions is well established. For Khyber Pass, the figures are typical of a congested traffic intersection in a major city, although none of the 80% of observations of NOx below 500 µg/m3 represent exceedances of the NO2 standard (which is 200 µg/m3 as a one-hour average).

Methods for NO2 estimates: deducing local concentrations plus background

Total NO2 concentrations are usually calculated as the sum of estimated NO2 from known local sources [NO2]est and background NO2 concentrations [NO2]bkd from all other sources:

[NO2] = [NO2]est + [NO2]bkd

The methods proposed below are used for estimating NO2 concentrations at a receptor due to a local NOx emission (ie, [NO2]est). Approaches to estimate background NO2 concentrations, or [NO2]bkd, have been discussed elsewhere (Ministry for the Environment, 2004a). For most of New Zealand, where background air is of rural or marine origin and has very small concentrations of NO2, the appropriate background value to use is zero. Where there are industries or traffic upwind but beyond the immediate area being modelled for NOx, there is the possibility of a higher 'background' concentration of NO2.

The total oxidation method

For the initial estimate, a total conversion of NO to NO2 is assumed:

[NO2]est = [NOx]mod

where [NOx]mod is the NOx concentration (µg/m3, expressed as NO2) at the receptor, observed or determined by a dispersion model, due to the known NOx emission under consideration. Thus if NOx is known or modelled, there is a method for estimating NO2.

The plots of observed NO2 and NOx concentrations in Figure A2.1 show that complete oxidation is always a very conservative assumption and will always result in significant overestimates of NO2 concentrations, especially where oxidation cannot be completed for more elevated NOx concentrations. The method gives a very conservative estimate of likely compliance with the NO2 standard and the possible number of exceedances. If adequate compliance without exceedances is indicated with this approach, no further estimates or calculations are necessary.

If the simple assumption of complete oxidation does not give confidence that compliance with the NO2 standard is adequate (eg, modelled or observed concentrations are close to, or exceeding, the NO2 standard), the practitioner should move to the ozone-limiting method outlined in the section below for lower (but still conservative) estimates of NO2 concentrations.

The ozone-limiting method

An approach based on the ozone-limiting method has been proposed to estimate NO2 concentrations from modelled NOx in New Zealand (Ministry for the Environment, 2004a).

The effect of ozone limiting is shown in the time series in Figure A2.2. Musick Point is about 10 km down the Tamaki River from the main industrial areas and motorways of South Auckland and is often downwind from them. The site experiences distinct diurnal changes in pollutant concentrations. A background condition is evident in Figure A2.2, with O3 about 60 µg/m3 and very small amounts of NOx (<<10 µg/m3). An assumption of a zero or very low NO2 background would be justified here. During pollution events, especially in late mornings, O3 is almost completely removed and replaced by NO2, and in some cases an excess of un‑reacted NO can be identified.

The usual approach (eg, Ministry for the Environment, 2004a) is modified here to estimate NO2 at a receptor. The estimate is then the sum of NO2 emitted at the local source plus the maximum concentration of NO2 that can be produced from NO using the available ozone:

[NO2]est = [NOx]mod x F(NO2) + 72

where F(NO2) is the mass fraction of NO2 in the NOx emission from the source under consideration, and the first term represents NO2 from the local source arriving un-oxidised at the receptor. Concentrations are all in µg/m3, expressed as NO2.

In the equation, 72 µg/m3 is the upper limit for NO2 formed by oxidation of NO by the maximum background O3 concentration. O3 concentrations in air coming off the ocean are quite predictable and show a seasonal variation, with the highest concentrations occurring during winter. At Baring Head near Wellington, NIWA has recorded maximum winter concentrations of about 35 ppb (75 µg/m3) and maximum summer concentrations of about 20 ppb (43 µg/m3). This maximum winter O3 concentration is sufficient to produce 72 µg/m3 of NO2 by oxidation of NO. When [NOx]mod < 72 µg/m3 then [NO2]est = [NOx]mod is used, as outlined in the total oxidation method above (ie, a total conversion of NO to NO2).

Figure A2.2: A time series of four winter days of NOx and O3 at Musick Point

See figure at its full size (including text description).

Validation at several monitoring sites in Auckland and Christchurch has shown that taking 10% NO2 in NOx emissions results in conservative NO2 estimates using the ozone-limiting method, particularly for elevated NOx concentrations, at most sites. An example of this is shown for Takapuna in Figure A2.3. The overestimate using the ozone-limiting method results from the combined effect of the actual percentage of NO2 in NOx being less than 10%, together with scavenging of both NO2 and O3 by vegetation and other surfaces, plus the establishment of the photochemical reaction, which eventually tends to reduce the NO2 concentration.

24- hour averages

When an assessment over 24-hour averaging periods is required it is advisable to use hourly data and rolling 24-hour averaging for the ozone-limiting method. This takes into account major diurnal fluctuations in the three main components of interest (NO2, NO and O3) and enables a meaningful result. Even in the sequence of pollutant events over several days shown in Figure A2.2, the 24-hour average never exceeds 20 µg/m3, although the corresponding ozone average remains between 35 and 50 µg/m3. Ozone-limiting assessments will thus still be conservative. This condition – fresh air cleaning out the results of pollution − is fortunately still common in New Zealand.

What next?

The methods above are useful where data for assessments are sparse and confidence is low. However, where there are well-defined emission data (maybe a small number of large sources) and sound meteorology, more precise modelling can be applied. The dispersion model Calpuff has a chemistry module that handles NOx transformations using default reaction rates and reagent ratios taken from the ambient ratio method. This will still be very conservative in the small windy cities of New Zealand but should provide good confidence in the assessment.

A realistic estimate of the fraction of NO2 in the emitted NOx is important to the success of the methods here. For most combustion sources (eg, power stations, vehicle emissions) the value of 10% NO2 in NOx emissions is usually considered a sufficiently conservative estimate.

Generally, the oxidation of NO to NO2 takes time to occur, during which time dispersion should have also occurred so that NO2:NOx ratios should not be too high (ie, less than 10%). There is, however, evidence (eg, Carslaw and Beevers, 2005) of NO2:NOx ratios in some vehicle emissions even higher than 20%. Roadside monitoring adjacent to traffic congestion (eg, Khyber Pass Road) frequently indicates high NOx concentrations with similarly high NO2:NOx ratios over 20% (see Figure A2.1). For this reason, sites such as this need more detailed monitoring and analysis than the simple methods outlined in this Appendix to assess compliance issues. Such monitoring and analysis can assist, in particular, with understanding the role of idling or slow-moving diesel engines and heavy vehicle engines under short-term loads (eg, accelerating uphill).

Summary

Two methods have been described here for estimating NO2 concentrations from modelled or measured NOx. The total oxidation method, assuming a total conversion of NO to NO2, always generates very conservative results, especially for elevated NOx concentrations. By assuming 10% NO2 in NOx emissions, the ozone-limiting method results in conservative NO2 concentrations at most sites. It should be noted that this method underestimates NO2 concentrations for some cases with nearby traffic sources of NOx (eg, at Khyber Pass Road), in which the NO2 proportion in NOx emissions may be higher than 10%. An improved estimate of the NO2 percentage in the NOx emission becomes important in these cases and it would be more appropriate to use site-specific NO2 and NOx monitoring data rather than the methods of estimation outlined here.

Where the ozone-limiting method also indicates the possibility of exceedances (eg, in the Khyber Pass record in Figure A2.1), a programme of real observations of NO2 is necessary to robustly determine the extent of the problem and monitor the success of measures taken to reduce it. Figure A2.4 is a flow diagram for the methods outlined.

Figure A2.3: Comparison between measured and modelled NO2 one-hour concentrations at Takapuna in 2001

See figure at its full size (including text description).

Figure A2.4: Summary of method for determining compliance with the NO2 standard

See figure at its full size (including text description).

[ |